A Prospective Study of Periodontal Disease and Pancreatic Cancer in US Male Health Professionals
Two previous cohort studies reported positive associations between tooth loss or periodontitis and pancreatic cancer risk. Data on periodontal disease were obtained at baseline and every other year thereafter in a cohort of 51 529 male health professionals aged 40–75 years. A total of 216 patients were diagnosed with incident pancreatic cancer during 16 years of follow-up. Multivariable relative risks (RRs) and 95% confidence intervals (CIs) were estimated using Cox proportional hazards models controlling for potential confounders, including detailed smoking history. All statistical tests were two-sided. Compared with no periodontal disease, history of periodontal disease was associated with increased pancreatic cancer risk (overall, multivariable RR = 1.64, 95% CI = 1.19 to 2.26; P = .002; crude incidence rates: 61 versus 25 per 100 000 person-years; among never smokers, multivariable RR = 2.09, 95% CI = 1.18 to 3.71; P = .01; crude incidence rates: 61 versus 19 per 100 000 person-years). In contrast, baseline number of natural teeth and cumulative tooth loss during follow-up were not strongly associated with pancreatic cancer. The association between periodontal disease and increased risk of pancreatic cancer may occur through plausible biologic mechanisms, but confirmation of this association is necessary.
Brief Communications 171
Affiliations of authors: Departments of
Epidemiology (DSM, KJ, EG) and Nutrition (EG),
Harvard School of Public Health, Boston, MA;
Channing Laboratory, Department of Medicine,
Brigham and Women’s Hospital and Harvard
Medical School, Boston, MA (DSM, EG, CSF);
School of Dentistry, Medical Sciences Campus,
University of Puerto Rico, San Juan, Puerto Rico
(KJ); Department of Oral Health Policy and
Epidemiology, Harvard School of Dental Medicine,
Boston, MA (KJ); Department of Adult Oncology,
Dana-Farber Cancer Institute, Boston, MA (CSF).
Correspondence to: Dominique S. Michaud, ScD,
Department of Epidemiology, Harvard School of
Public Health, Kresge 920, 677 Huntington Ave.,
Boston, MA 02115 (e-mail: dmichaud@hsph.
See “ Notes ” following “ References. ”
© The Author 2007. Published by Oxford University
Press. All rights reserved. For Permissions, please
Cancer of the pancreas, the fourth leading
cause of cancer death in the United States,
is a rapidly fatal malignancy ( 1 ). The only
established modifiable risk factor for pan-
creatic cancer is cigarette smoking, although
data suggest that diabetes, obesity, and
insulin resistance are also associated with
risk ( 2 – 5 ). In addition, chronic pancreatitis
has been associated with extremely high
risks of pancreatic cancer ( 6 ), suggesting
that inflammation may be involved in the
initiation and/or promotion of pancreatic
cancer. Inflammation may enhance cellular
proliferation and mutagenesis, reduce
adaptation to oxidative stress, promote
angiogenesis, inhibit apoptosis, and increase
secretion of inflammatory mediators ( 7 ).
Serum levels of C-reactive protein and
of other biomarkers of systemic inﬂ amma-
tion are consistently higher in individuals
with periodontal disease than in those
who have no periodontal disease ( 8 , 9 ).
Periodontitis is the primary cause of tooth
loss in adults and develops over many years
as a result of bacterial infection and inﬂ am-
mation of the gum that spread to the liga-
ments and bone that support the teeth.
Positive associations between tooth loss
cancer, were obtained on the baseline ques-
tionnaire and on subsequent biennial ques-
tionnaires. Deaths of most members of this
cohort were reported by family members
or by the postal service in response to ques-
tionnaire mailings. In addition, the National
Death Index was searched biennially for
nonrespondents; this method has been
shown to have a sensitivity of 98% ( 12 ).
Most diagnoses of pancreatic cancer
(reported or detected through the death
follow-up) were conﬁ rmed with medical
records (95%). Remaining self-reported
cancers were conﬁ rmed by a secondary
source (e.g., death certiﬁ cate, physician, or
telephone interview of a family member).
This study was approved by the Human
Subjects Committee of the Harvard School
of Public Health. Written informed con-
sent was assumed on completion of the
On the baseline questionnaire, partici-
pants responded to the following question:
“ Have you had periodontal disease with
bone loss? ” This question was validated
among dentists ( 13 ) and nondentists ( 14 ) in
the HPFS cohort by obtaining radiographs
from subsets of individuals with and without
a self-reported history of periodontal dis-
ease. Among dentists, the positive and neg-
ative predictive values were 0.76 and 0.74,
respectively ( 13 ). Among nondentists, the
positive and negative predictive values were
0.80 and 0.68, respectively ( 14 ).
Participants reported number of natural
teeth at baseline, and any tooth loss during
the past 2 years was reported biennially on
A Prospective Study of Periodontal
Disease and Pancreatic Cancer in US
Male Health Professionals
Dominique S. Michaud , Kaumudi Joshipura , Edward Giovannucci ,
Charles S. Fuchs
Two previous cohort studies reported positive associations between tooth loss or
periodontitis and pancreatic cancer risk. Data on periodontal disease were obtained
at baseline and every other year thereafter in a cohort of 51 529 male health profes-
sionals aged 40 – 75 years. A total of 216 patients were diagnosed with incident pan-
creatic cancer during 16 years of follow-up. Multivariable relative risks (RRs) and
95% confidence intervals (CIs) were estimated using Cox proportional hazards mod-
els controlling for potential confounders, including detailed smoking history. All
statistical tests were two-sided. Compared with no periodontal disease, history of
periodontal disease was associated with increased pancreatic cancer risk (overall,
multivariable RR = 1.64, 95% CI = 1.19 to 2.26; P = .002; crude incidence rates: 61
versus 25 per 100 000 person-years; among never smokers, multivariable RR = 2.09,
95% CI = 1.18 to 3.71; P = .01; crude incidence rates: 61 versus 19 per 100 000
person-years). In contrast, baseline number of natural teeth and cumulative tooth
loss during follow-up were not strongly associated with pancreatic cancer. The
association between periodontal disease and increased risk of pancreatic cancer
may occur through plausible biologic mechanisms, but confirmation of this associa-
tion is necessary.
J Natl Cancer Inst 2007;99: 171 – 5
and pancreatic cancer were reported in two
recent studies ( 10 , 11 ). Data from these
studies were obtained prospectively,
thereby ruling out potential recall bias, but
residual confounding by amount of smok-
ing remains a possible explanation, given
that all participants were smokers in one
study ( 10 ), and no adjustments for smoking
were presented in the analyses of pancre-
atic cancer in the other ( 11 ).
We therefore examined the relationship
between periodontitis and tooth loss and
the subsequent risk of pancreatic cancer
in a large prospective cohort of health
pro fessionals. This cohort provides a
homo geneous population with respect to
socioeconomic status, and detailed data on
smoking history allow for tight adjustments
The Health Professionals Follow-Up
Study (HPFS) was initiated in 1986 and
included 51 529 predominantly white US
men aged 40 – 75 years. Participants in this
cohort are dentists (57.6%), veterinarians
(19.6%), pharmacists (8.1%), optometrists
(7.3%), osteopathic physicians (4.3%), and
podiatrists (3.1%). Individual data on life-
style and medical conditions, including
172 Brief Communications
JNCI Vol. 99, Issue 2
January 17, 2007
follow-up questionnaires. Self-reported
number of teeth is highly correlated with
the actual number of teeth on clinical
examination ( r = .97) ( 15 ). Accordingly,
we expect self-reported number of teeth
and tooth loss during follow-up to be accu-
rate in this cohort of educated health
In the main analyses, 48 375 men were
eligible after excluding participants who
were diagnosed with cancer (other than
nonmelanoma skin cancer) before 1986 and
those with missing data on periodontal dis-
ease (N = 1118). After exclusions, a total of
216 conﬁ rmed incident pancreatic cancer
case patients were available for analyses.
For the incident tooth loss analyses, with
follow-up starting in 1988, 190 case patients
We computed person-time of follow-up
for each participant from the return date of
the baseline questionnaire to the date of
CONTEXT AND CAVEATS
Tooth loss or periodontal disease, inflam-
mation of the gums caused by bacterial
infection that leads to tooth loss, was
shown to be associated with increased risk
of pancreatic cancer in two studies. In one
study, all subjects were smokers, and in the
other, no adjustment was made for smok-
ing, a known risk factor for pancreatic
Questionnaire-based prospective study of
pancreatic cancer incidence and history of
periodontal disease among male health
professionals that includes adjustment for
History of periodontal disease was inde-
pendently associated with an increased risk
of pancreatic cancer overall and in never
smokers; recent tooth loss was associated
with additional increased risk.
The association may be due to systemic
inflammation and/or increased levels of
carcinogenic compounds generated by bac-
teria in the oral cavity of individuals with
History of periodontal disease was self-
reported and may be subject to measure-
pancreatic cancer diagnosis, death from any
cause, or the end of follow-up (January 31,
2002), whichever came ﬁ rst. Relative risks
(RRs) adjusted for potential confounders
and their 95% conﬁ dence intervals (CIs)
were approximated by Cox proportional
hazards regression. For the periodontal
disease analyses, baseline data on periodon-
tal disease were used without updating
because periodontal disease is chronic and
generally progresses slowly. However, in a
secondary analysis, we updated periodontal
disease over each 2-year cycle. For the
analyses of tooth loss, we examined the
effect of tooth loss during the study period
because remote tooth loss is more likely
to reﬂ ect caries than periodontal disease.
For the incident tooth loss analyses, follow-
up started in 1988 because the baseline
questionnaire did not assess recent tooth
loss. We calculated cumulative incident
tooth loss for our analysis. Details on anal-
yses and power calculations are available
online as Supplementary Data.
Models were stratiﬁ ed by age and calen-
dar time (i.e., time metameter in Cox pro-
portional hazards models) and adjusted for
cigarette smoking (time-varying) ( 16 ), dia-
betes (cumulative updating), baseline body
mass index (BMI) ( 17 ), race (African
American/other), height, physical activity,
nonsteroidal anti-inﬂ ammatory drug use
( 18 ), multivitamin use, geographic regions
(to serve as vitamin D proxy), profession
(dentist versus other), and history of chole-
cystectomy (cumulative updating) ( 19 ). In
addition, we controlled for a number of
dietary factors, including baseline intakes
of vitamin D, calcium, sucrose, fruits and
vegetables, and total calories. Alcohol was
not included because it is not related to
pancreatic cancer in this cohort ( 20 ), and
including it in the multivariable model did
not change the estimates. The Cox propor-
tional hazards models all satisﬁ ed the pro-
portionality of hazards assumption. In the
tables, models with the dietary covariates
are shown separately because 1546 individ-
uals, including eight patients, with inade-
quate dietary data were removed from the
analysis. All P values (two-sided) were cal-
culated using a chi-square test. All statisti-
cal procedures were performed using SAS
release 9.1 (SAS Institute, Cary, NC).
Participants with periodontal disease
were older, more likely to have a history of
diabetes, and more likely to be current
smokers than those who had no reported
history of periodontal disease at baseline
(Supplementary Table 1, available online).
Other potential pancreatic cancer risk
factors, including BMI, were similar across
periodontal disease status.
History of periodontal disease was asso-
ciated with increased risk for pancreatic
cancer, after adjusting for age ( Table 1 ).
The association was attenuated in a multi-
variable model adjusting for potential con-
founders (smoking accounted for most of
the attenuation in risk) but remained statis-
tically signiﬁ cant ( Table 1 ). Further adjust-
ment for a number of dietary variables did
not affect this association (RR = 1.64, 95%
CI = 1.19 to 2.26; P = .002; crude incidence
rates: 61 versus 25 per 100 000 person-
years; Table 1 ). In a secondary analysis, we
updated periodontal disease status through-
out the follow-up period and observed a
similar, albeit slightly weaker, association
for pancreatic cancer (multivariable RR =
1.52, 95% CI = 1.13 to 2.05).
Number of natural teeth at baseline was
not associated with pancreatic cancer
( Table 1 ). However, cumulative incident
tooth loss during follow-up (1988 – 2002),
which was modeled as a time-varying expo-
sure in a Cox proportional hazards model,
was associated with a statistically nonsignif-
icant increase in pancreatic cancer ( Table 1 ).
When periodontal disease was added to the
model, the association between incident
tooth loss and pancreatic cancer was fur-
ther attenuated, suggesting that tooth loss
may not be an independent risk factor
( Table 1 ). Tooth loss during the past 4
years was more strongly associated with
pancreatic cancer than cumulative tooth
loss ( Table 1 ). This association was also
attenuated after controlling for periodontal
disease at baseline but remained statistically
signiﬁ cant ( Table 1 ). In this model, the
association between periodontal disease
and pancreatic cancer was similar (RR =
1.58, 95% CI = 1.13 to 2.22; P = .007).
In a joint analysis, periodontal disease
with recent tooth loss was associated with a
2.7-fold increase in pancreatic cancer, com-
pared with no periodontal disease and no
recent tooth loss ( Table 2 ). This joint
effect suggests that timing and severity of
periodontal disease, as reﬂ ected by those
who experience recent tooth loss, are
important factors in this association, and
periodontal disease may increase risk
Brief Communications 173
through promotion of initiated cells.
Alternatively, the joint classiﬁ cation of
recent tooth loss and periodontal disease
may improve the classiﬁ cation of periodon-
tal disease, thereby reducing measurement
error and reﬁ ning the association.
We conducted stratiﬁ ed analyses to
assess the possibility of any residual con-
founding by known pancreatic cancer risk
factors ( Table 3 ). The association between
periodontal disease and pancreatic cancer
appeared similar among younger and older
participants, suggesting that age is unlikely
to be confounding the overall association.
Moreover, the inﬂ uence of periodontal
disease appeared to be stronger among
men who were never smokers (RR = 2.09,
95% CI = 1.18 to 3.71; P = .01; crude inci-
dence rates: 61 versus 19 per 100 000 per-
son-years) or among men with BMI less
than 25 kg/m
(RR = 2.20, 95% CI = 1.34
to 3.61; P = .002), suggesting that smoking
or obesity is unlikely to explain the associa-
tion. In addition, when men with a history
of diabetes were excluded from the analy-
sis, a statistically signiﬁ cant association
for periodontal disease was still observed
Table 1. Periodontal disease, tooth loss, and baseline teeth number and risk of pancreatic cancer in the Health Professionals Follow-up
Study cohort *
Characteristic N Person-years
RR (95% CI)
RR (95% CI) †
RR (95% CI) ‡
RR (95% CI) §
1986 – 2002)
No 149 590 844 1.0 (referent) 1.0 (referent) 1.0 (referent)
Yes 67 109 184 1.83 (1.36 to 2.45) 1.63 (1.19 to 2.22) 1.64 (1.19 to 2.26)
No. of natural teeth
1986 – 2002)
25 – 32 160 588 527 1.0 (referent) 1.0 (referent) 1.0 (referent)
17 – 24 37 77 308 1.20 (0.83 to 1.73) 1.09 (0.75 to 1.58) 1.00 (0.67 to 1.48)
0 – 16 19 34 055 1.22 (0.75 to 1.99) 1.02 (0.61 to 1.69) 1.02 (0.61 to 1.71)
Updated tooth loss
1988 – 2002) ||
None 130 465 498 1.0 (referent) 1.0 (referent) 1.0 (referent) 1.0 (referent)
Any 60 107 646 1.40 (1.02 to 1.93) 1.31 (0.94 to 1.84) 1.31 (0.92 to 1.85) 1.21 (0.86 to 1.71)
Incident in the
past 4 y
None 139 499 779 1.0 (referent) 1.0 (referent) 1.0 (referent) 1.0 (referent)
Any 51 73 365 1.73 (1.24 to 2.41) 1.63 (1.16 to 2.31) 1.61 (1.13 to 2.31) 1.53 (1.08 to 2.17)
* RR = relative risk; CI = confidence interval.
† Adjusted for age, smoking history (never; quit ≥ 15 years; quit <15 years, ≤ 25 pack-years; quit <15 years, >25 pack-years; current ≤ 25 pack-years; current >25
pack-years), profession (dentist/others), race (African American/other), geographic location (northeast, south, midwest, west), physical activity (quintiles), history
of diabetes (yes/no), body mass index (quintiles), height (quintiles), history of cholecystectomy (yes/no), nonsteroidal anti-inflammatory drug use (yes/no), multivi-
tamin use (yes/no), and in addition, for periodontal and tooth loss models, baseline teeth number (0 – 16, 17 – 24, 25 – 32).
‡ Additionally controlling for dietary intakes (quintiles) of fruits and vegetables, vitamin D, calcium, sucrose, and total calories. P = .002 for periodontal disease at
baseline, P = .009 for tooth loss in past 4 years, calculated using two-sided chi-square tests.
§ Adjusting for periodontal disease in addition to other covariates in model ( † ) above. P = .02 for tooth loss in the past 4 years, calculated using a two-sided
|| Excludes edentulous participants (as reported in 1986). Follow-up for this analysis starts in 1988 and includes 190 cases.
Table 2. History of periodontal disease, incident tooth loss, and risk of pancreatic cancer
in the Health Professionals Follow-up Study (1988 – 2002) *
Periodontal disease † N Person-years Multivariable RR (95% CI) ‡ P §
No tooth loss in past 4 y 106 435 031 1.0 (referent)
Tooth loss in past 4 y 24 50 818 1.29 (0.81 to 2.06) .28
No tooth loss in past 4 y 33 64 748 1.38 (0.92 to 2.09) .12
Tooth loss in past 4 y 27 22 547 2.71 (1.70 to 4.32) <.001
* Excludes edentulous participants (as reported in 1986). Follow-up for this analysis starts in 1988 and
includes 190 patients. RR = relative risk; CI = confidence interval.
† As reported at baseline (1986).
‡ Adjusted for age, smoking history (never; quit ≥ 15 years; quit <15 years, ≤ 25 pack-years; quit <15 years,
>25 pack-years; current ≤ 25 pack-years; current >25 pack-years), profession (dentist/others), race (African
American/other), geographic location (northeast, south, midwest, west), physical activity (quintiles),
history of diabetes (yes/no), body mass index (quintiles), height (quintiles), history of cholecystectomy
(yes/no), nonsteroidal anti-inflammatory drug use (yes/no), multivitamin use (yes/no), and baseline teeth
number (0 – 16, 17 – 24, 25 – 32).
§ P values were calculated using a two-sided chi-square test.
174 Brief Communications
JNCI Vol. 99, Issue 2
January 17, 2007
(RR = 1.70, 95% CI = 1.23 to 2.35;
P = .001, n = 191 cases).
Because dentists may be less likely to
underreport periodontal disease with bone
loss (as asked on the questionnaire) than
other health professionals, we assessed asso-
ciations separately by healthcare profession.
We observed a stronger association among
dentists than among other health profession-
als in this cohort ( Table 3 ). The accuracy of
self-reported periodontal disease was slightly
higher among dentists than for other health
professionals in validation studies of self-
reported history of periodontal disease in
this cohort ( 13 , 14 ). Therefore, undifferential
measurement error from potential underre-
porting of periodontitis in the nondentist
group may explain the weaker association in
that group. A similar difference between
dentists and nondentists was observed for
periodontal disease and ischemic stroke in
this cohort (among dentists, RR = 1.46, 95%
CI = 1.07 to 2.00, versus among nondentists,
RR = 1.13, 95% CI = 0.73 to 1.72) ( 21 ). The
prevalence of periodontal disease in this
study (16%) is similar to the prevalence in
US adults aged 30 – 90 years (13% for
moderate or severe form of periodontitis
based on National Health and Nutrition
Examination Survey III data) ( 22 ).
Although individuals with periodontal
disease may change their diet as their gums
become more sensitive or as they start los-
ing teeth, controlling for a variety of dietary
factors had no effect on the relative risks.
Finally, because periodontal disease and
Table 3. Periodontal disease and risk of pancreatic cancer in the Health Professionals
Follow-up Study cohort 1986 – 2002, by smoking status, body mass index (BMI), age, and
disease N Person-years
RR (95% CI) ‡ P ‡
No 55 288 269 1.0 (referent)
Yes 19 31 159 2.09 (1.18 to 3.71) .01
No 87 279 452 1.0 (referent)
Yes 47 73 632 1.63 (1.12 to 2.37) .01
No 53 270 320 1.0 (referent)
Yes 29 46 768 2.20 (1.34 to 3.61) .002
No 92 306 661 1.0 (referent)
Yes 35 58 998 1.32 (0.87 to 2.00) .19
No 49 405 727 1.0 (referent)
Yes 19 57 094 1.85 (1.06 to 3.24) .03
No 100 185 117 1.0 (referent)
Yes 48 52 092 1.59 (1.11 to 2.29) .01
No 83 333 443 1.0 (referent)
Yes 51 73 842 1.91 (1.31 to 2.78) <.001
No 66 257 401 1.0 (referent)
Yes 16 35 343 1.17 (0.65 to 2.11) .59
* Seven patients missing BMI; eight missing smoking status.
† RR = relative risk; CI = confidence interval. Adjusted for age, race (African American/other), geographic
location (northeast, south, midwest, west), physical activity (quintiles), history of diabetes (yes/no), BMI
(quintiles), height (quintiles), history of cholecystectomy (yes/no), nonsteroidal anti-inflammatory drug use
(yes/no), multivitamin use (yes/no), profession (dentist/others, except in models stratified by profession),
and smoking history (never; quit ≥ 15 years; quit <15 years, ≤ 25 pack-years; quit <15 years, >25 pack-
years; current ≤ 25 pack-years; current > 25 pack-years, except in model of never smokers).
‡ P values were calculated using a two-sided chi-square test.
tooth loss were assessed before any diagno-
sis of pancreatic cancer, recall and selection
biases were minimized, and thus the need
for proxy interviews was eliminated.
Several mechanisms could potentially
explain the observations from this study.
Inﬂ ammation appears to play an important
role in pancreatic cancer pathogenesis ( 23 ),
although the inﬂ ammatory mediators that
lead to the development of pancreatic can-
cer remain poorly deﬁ ned. An association
between periodontal disease and systemic
inﬂ ammation has been observed using bio-
markers ( 8 ). In the HPFS, plasma C-
reactive protein levels were 30% higher in
individuals with a history of periodontal
disease than in those with no history ( 9 ).
We hypothesize that periodontal disease
may promote pancreatic carcinogenesis
through inﬂ ammation.
Alternatively, periodontal disease could
inﬂ uence pancreatic carcinogenesis through
increased generation of carcinogens,
namely nitrosamines ( 10 , 24 ). Individuals
with periodontal disease and poor oral
hygiene have elevated levels of oral bacteria
and have much higher nitrosamine levels in
their oral cavity ( 25 ) due to nitrate- reducing
bacteria ( 26 ). Nitrosamines and gastric
acidity have been hypothesized to have an
important role in pancreatic cancer; numer-
ous studies support this hypothesis ( 27 ).
In this study, history of periodontal dis-
ease was self-reported, and measurement
error due to inaccurate reporting could
have occurred; this error would be nondif-
ferential, however, and would produce
attenuation toward the null given the binary
exposure, thereby weakening the underly-
ing association. Although our population
is fairly homogeneous, we did not have
income level or other socioeconomic status
data on participants in this cohort and can-
not additionally adjust for these factors; if
an unknown risk factor of pancreatic cancer
was associated with socioeconomic status,
it may have confounded our association.
In summary, we observed an association
between history of periodontal disease and
risk of pancreatic cancer that was indepen-
dent of other known or suspected risk fac-
tors for this malignancy. Moreover,
increased severity of periodontal disease, as
manifested by periodontitis with recent
tooth loss, was associated with the greatest
risk. Given our limited understanding of
pancreatic cancer etiology, we believe that
Brief Communications 175
further investigation into this relation and
the role of systemic inﬂ ammation in pancre-
atic carcinogenesis is warranted.
(1) American Cancer Society . Cancer facts &
ﬁ gures 2006 . Atlanta (GA) : American Cancer
Society, Inc. ; 2006 .
(2) Everhart J , Wright D . Diabetes mellitus as
a risk factor for pancreatic cancer. A meta-
analysis . JAMA 1995 ; 273 : 1605 – 9 .
(3) Michaud DS . The epidemiology of pancre-
atic, gallbladder, and other biliary tract
cancers . Gastrointest Endosc 2002 ; 56 Suppl 2:
S195 – 200 .
(4) Jee SH , Ohrr H , Sull JW , Yun JE , Ji M ,
Samet JM . Fasting serum glucose level and
cancer risk in Korean men and women . JAMA
2005 ; 293 : 194 – 202 .
(5) Stolzenberg-Solomon RZ , Graubard BI ,
Chari S , Limburg P , Taylor PR , Virtamo J ,
et al . Insulin, glucose, insulin resistance, and
pancreatic cancer in male smokers . JAMA
2005 ; 294 : 2872 – 8 .
(6) Lowenfels AB , Maisonneuve P , Lankisch PG .
Chronic pancreatitis and other risk factors for
pancreatic cancer . Gastroenterol Clin North
Am 1999 ; 28 : 673 – 85, x .
(7) Coussens LM , Werb Z . Inﬂ ammation and
cancer . Nature 2002 ; 420 : 860 – 7 .
(8) Loos BG . Systemic markers of inﬂ ammation
in periodontitis . J Periodontol 2005 ; 76 (Suppl) :
2106 – 15 .
(9) Joshipura KJ , Wand HC , Merchant AT ,
Rimm EB . Periodontal disease and biomark-
ers related to cardiovascular disease . J Dent
Res 2004 ; 83 : 151 – 5 .
(10) Stolzenberg-Solomon RZ , Dodd KW ,
Blaser MJ , Virtamo J , Taylor PR , Albanes D .
Tooth loss, pancreatic cancer, and Helicobacter
pylori . Am J Clin Nutr 2003 ; 78 : 176 – 81 .
(11) Hujoel PP , Drangsholt M , Spiekerman C , Weiss
NS . An exploration of the periodontitis -cancer
association . Ann Epidemiol 2003 ; 13 : 312 – 6 .
(12) Rich-Edwards JW , Corsano KA , Stampfer MJ .
Test of the National Death Index and Equifax
Nationwide Death Search . Am J Epidemiol
1994 ; 140 : 1016 – 9 .
(13) Joshipura KJ , Douglass CW , Garcia RI ,
Valachovic R , Willett WC . Validity of a self-
reported periodontal disease measure . J Public
Health Dent 1996 ; 56 : 205 – 12 .
(14) Joshipura KJ , Pitiphat W , Douglass CW .
Validation of self-reported periodontal mea-
sures among health professionals . J Public
Health Dent 2002 ; 62 : 115 – 21 .
(15) Douglass CW , Berlin J , Tennstedt S . The
validity of self-reported oral health status in the
elderly . J Public Health Dent 1991 ; 51 : 220 – 2 .
(16) Fuchs C , Colditz G , Stampfer M , Giovannucci
E , Hunter D , Rimm E , et al . A prospective
study of cigarette smoking and the risk of
pancreatic cancer . Arch Intern Med 1996 ; 156 :
2255 – 60 .
(17) Michaud DS , Giovannucci E , Willett WC ,
Colditz GA , Stampfer MJ , Fuchs CS . Physical
activity, obesity, height and the risk of pancre-
atic cancer . JAMA 2001 ; 286 : 921 – 9 .
(18) Schernhammer ES , Kang JH , Chan AT ,
Michaud DS , Skinner HG , Giovannucci E ,
et al . A prospective study of aspirin use and
the risk of pancreatic cancer in women . J Natl
Cancer Inst 2004 ; 96 : 22 – 8 .
(19) Schernhammer ES , Michaud DS , Leitzmann
MF , Giovannucci E , Colditz GA , Fuchs CS .
Gallstones, cholecystectomy, and the risk for
developing pancreatic cancer . Br J Cancer
2002 ; 86 : 1081 – 4 .
(20) Michaud DS , Giovannucci E , Willett WC ,
Colditz GA , Fuchs CS . Coffee and alcohol
consumption and the risk of pancreatic cancer
in two prospective United States cohorts .
Cancer Epidemiol Biomarkers Prev 2001 ; 10 :
429 – 37 .
(21) Joshipura KJ , Hung HC , Rimm EB ,
Willett WC , Ascherio A . Periodontal disease,
tooth loss, and incidence of ischemic stroke .
Stroke 2003 ; 34 : 47 – 52 .
(22) Albandar JM , Brunelle JA , Kingman A .
Destructive periodontal disease in adults 30
years of age and older in the United States,
1988 – 1994 . J Periodontol 1999 ; 70 : 13 – 29 .
(23) Farrow B , Evers BM . Inﬂ ammation and the
development of pancreatic cancer . Surg Oncol
2002 ; 10 : 153 – 69 .
(24) Abnet CC , Kamangar F , Dawsey SM ,
Stolzenberg-Solomon RZ , Albanes D ,
Pietinen P , et al . Tooth loss is associated with
increased risk of gastric non-cardia adenocar-
cinoma in a cohort of Finnish smokers . Scand
J Gastroenterol 2005 ; 40 : 681 – 7 .
(25) Nair J , Ohshima H , Nair UJ , Bartsch H .
Endogenous formation of nitrosamines and
oxidative DNA-damaging agents in tobacco
users . Crit Rev Toxicol 1996 ; 26 : 149 – 61 .
(26) Shapiro KB , Hotchkiss JH , Roe DA .
Quantitative relationship between oral nitrate-
reducing activity and the endogenous forma-
tion of N-nitrosoamino acids in humans .
Food Chem Toxicol 1991 ; 29 : 751 – 5 .
(27) Risch HA . Etiology of pancreatic cancer, with
a hypothesis concerning the role of N-nitroso
compounds and excess gastric acidity . J Natl
Cancer Inst 2003 ; 95 : 948 – 60 .
Dr D. S. Michaud had full access to all of the
data in the study and takes responsibility for the
integrity of the data and the accuracy of the data
The work reported in this manuscript was sup-
ported by CA55075 and CA95589.
We thank Dr Walter Willett for his valuable
advice, support, and dedication to the success of
Manuscript received July 25 , 2006 ; revised
October 27 , 2006 ; accepted October 30 , 2006.